Dear William,
>
> I still get the same error message when I try to use "--subsamp=8,4,2,2"
> even if I also use "--warpres=10,10,12". I.e. "Exception thrown with
> message: St9bad_alloc". Is there something else I should try?
Hmm, there is no known bug that can explain that. Could you please upload
a copy of the data you are using along with the command line and any
config-files you might have edited? Go to
http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi and follow the instructions
there, and then email me the six digit code that you are given.
> very high warp resolution in the z-direction, will that save any execution
> time by constraining fnirt's search in that direction? Or will these
> dimensional constraints have to wait for the new version of fnirt that you
> mention is in development?
It is a little bit more tricky I am afraid. The warps need to be
uni-dimensional in the space of the EPI (the --in image). With your
suggestion they will be uni-dimensional in the --ref space, which means
that any initial affine transform will take those "uni-dimensional arrows"
and rotate them. It is of course still fundamentally 1D, but those
rotations will need to be taken care of.
Good Luck Jesper
|